P. Branchini (INFN Roma 3) Involved Group: INFN-LNF G. Felici, INFN-NA A. Aloisio, INFN-Roma1 V. Bocci, INFN-Roma3

Transcription

1 P. Branchini (INFN Roma 3) Involved Group: INFN-LNF G. Felici, INFN-NA A. Aloisio, INFN-Roma1 V. Bocci, INFN-Roma3

2 Let s remember the specs in SuperB Baseline: re-implement BaBar L1 trigger with some improvements Shorter latency (~6us instead of 12us) Higher sampling frequencies (DCH and EMC) 2-d map for calorimeter Possible additions SVT trigger What about the TOF? (in DELPHI we used it in the trigger) Bhabha Veto Do we need an absolute time stamp at the trigger level? Challenge To keep the event loss due to dead time below 1% => a maximum of ~70ns per-event dead time is allowed in trigger and FCTS Other considerations What goes in L1, what in L3, what s the optimum? Luitz 2010 P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

3 Triggered event apple pie Both 40% Triggered by EMC 10% Triggered by DC 50% The original BaBar trigger time resolution was about 100 ns. Can we improve it? 100 ns good for BaBar μs dead time. We should aim at 30 ns since SuperB dead time is estimated to be 70 ns. P. D. Dauncey et Al., Design and performance of the level 1 calorimeter trigger for the BABAR detector (2001) P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

4 BaBar: Trigger Layout J. C. Andress et Al., BaBar Calorimeter Level 1 Trigger Design, BaBar Note (1998) P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

5 Latency in DC Latency in BaBar : DC event processing about 6 μs + GLT event processing 3 μs + trigger data movement 3,8 μs Total: 12.8 μs How to reach about 6 μs? Latency of the system is dominated by DC processing. It depends on how fast data processing is. We should be able to be faster with new FPGAs. P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

6 SuperB DC trigger Following BaBar strategy we would like to implement fast tracking but why? For background reduction in L3 trigger(see picture below) For event classification Real time data quality check and detector monitoring.. Z0 distrib. of triggered events before zed measurement upgrade P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

7 P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

8 Track finder P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

9 DC track finder algorithm Faster sampling means better precision (if the gas mixture is the same) do we need it? Chai 2004 P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

10 track finder algorithm track finder algorithm P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

11 What we get in the B-> D* K sample Zed resolution RMS=6.5 cm R resolution RMS = 2.4 cm Hypothesis: error on x coordinate = error on y coordinate : gaussian, sigma 0.8 mm error on z coordinate : gaussian, sigma = 1.5 cm P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

12

13

14 DC daq and trigger paths A dedicated path for the trigger is already foreseen for the DC. P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

15 Trigger Path Thr Clk LUT P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

16 dedicated trigger path The information from the fast path feeds a 1 bit flash adc (a prototype is being drawn by Roma3). The sampling frequency should be optimised Possible range:(from 3.5 to 56 MHz). In the prototype we would like to be able to vary the sampling frequency. We would like to test prototype and strategy on the LNF DC prototype. These bits (one for each wire) are delivered to a FPGAs via LVDS links (EMC like). P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

17 A possible sketch of trigger system in SuperB DC and EMC trigger crates have a common inteface (LVDS or optical) with pertaining sub-detectors. EMC(i) and DC(i) boards share a common hw platform and only differ in firmware. P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

18 Do we need rad hard equipment? I m not completely convinced. We could extensively make use of protocol correction error, thus avoiding to use real rad hard equipment if not strictly necessary. Moreover we could also think to house the most critical part of the radiation sensitive hardware in a shielded zone. P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

19 For instance P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

20 Thank you for your kind attention! P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

21 Latency in the EMC Barrel ADC conversion time. Depends on Shaping, can it be changed? J. C. Andress et Al., Babar Calorimeter Level 1 Trigger Design, BaBar Note (1998) Can we improve latency here? P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

22 Forward N N cryst Towers Barrel N N cryst Towers = 25 = 176 = 12 EMC fast trigger path and energy one = 480 Crystal n Energy accurate measurement path CSP Double Range shaper X N cryst V peak = if ( Range, K HREn, K LREn) Range (0,1) n n Trigger Fast Path Analog Pulse Encoder T width = K t n= N cryst n= 1 LVDS E n T = T 0 + T fix + T jitter P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

23 EMC forward tentative implementation X5 Double Range shaper Range CSP X5 X5 5 input VFE VME Range Shaper New Circuit Analog Sum over 5 input. 5 input Analog Pulse Encoder LVDS T = T 0 + T fix + T jitter Trigger Tower module P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

24 Barrel CsI (Tl doped) original BaBar P. D. Dauncey et Al., Design and performance of the level 1 calorimeter trigger for the BABAR detector (2001) 3 thresholds applied to the signal of the tower. The plot shows the difference between the EMC trigger time and the DC trigger time. μ+μ- events used to select good A FIR Filter with 8 parameters was applied to the signal. resolution Its zero in DC. crossing occurred at roughly a fixed time distance from the start of the signal, it was used to gate the threshold information. Due to this mechanism the time resolution was about 100 ns. P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

25 Forward Lyso signal The Lyso has been read out by a PMT (signal lasts 150 ns) we should not have problems here. Moreover a resolution on 1 ns has been reached on the peak charge distribution at CERN test beam. P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

26 SuperB EMC EMC Barrel : 5760 CsI(Tl) Crystals EMC Forward = 4400 Lyso Crystals (176 modules) P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

27 Comments I The afore mentioned strategy worked fine for BaBar. Unfortunately latency is dominated by adc conversion and peak finding. Peak finding determined time jiitter too. We really should try to improve this part. P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June

28 Comments II At the moment we could immagine to assert a trigger signal using only a part of the energy deposit in the CsI. Resolution could be worse by the sqrt(energy fraction collected). But we do not have any crystal to play with. We are waiting for a small sample of BaBar crystals From Bill P. Branchini- INFN Roma3 - SuperB Workshop (ELBA) 28 May 2 June